The occurrence of pediatric intracranial germ cell tumors (PIGCTs) is significantly higher in Asia than in the West.1,2 Survival of patients with PIGCTs is mainly determined by their histology types.3 The 10-year survival rate of patients with pure germinoma can be as high as 92.7%. In contrast, patients with certain types of germ cell tumors (yolk sac tumor, embryonal carcinoma, or choriocarcinoma) have 3-year survival rates as low as 27.3%.4
Intracranial tumoral bleeding is rare in pediatric patients. Patients with choriocarcinoma and nongerminomatous germ cell tumors (NGGCTs) with choriocarcinoma elements are notorious for being at high risk of tumoral hemorrhage (TH) and poor survival.5 However, the influence of TH on functional and survival outcomes of PIGCTs has not been well studied.2 The purpose of this study was to identify the factors affecting functional and survival outcomes of PIGCT patients with TH. To achieve this goal, we evaluated clinical and magnetic resonance imaging (MRI) findings in these patients. By more aggressively managing these factors, we would expect to improve the long-term prognosis of these pediatric patients.
The protocol of this retrospective study was approved by our institutional review board, giving us access to the clinical records and radiological images of the patients with PIGCTs.
2.1. Patients and clinical findings
There were 116 pediatric patients diagnosed with intracranial germ cell tumors in our institute between 2002 and 2016. Seventeen of these patients also had TH at the time of their initial diagnosis. TH was diagnosed by pathological examination in 13 patients and by clinical and MR results in 4 patients. Clinical findings, including demographic features, chief clinical symptoms, final diagnosis, and functional and survival outcomes of treatment, were evaluated (Table 1). Any recurrence of diseases or complications of treatment were recorded. The modified Rankin scale (mRS) was used to assess functional outcome. “Poor outcome” was defined as mRS ≧ 3, and “favorable outcome” was defined as mRS ≦ 2.6
2.2. MRI findings
The MR images of the brain and spine were acquired on 1.5T MR scanners (Signa HD & Excite TwinSpeed; GE Healthcare, Waukesha, WI). The sequences included a T1-weighted image, FLAIR, T2-weighted image, and 3D-contrast-enhanced T1-weighted image. Diffusion MR images with b = 0 and b = 1000 s/mm2 were available for 14 patients. All patients also underwent contrast-enhanced spinal MRI to elucidate whether any seeding by the tumor had occurred during the initial diagnosis and subsequent follow-up period. The intervals between acute symptoms onset and the time of the MR image examination were recorded. The pretreatment scans were analyzed. The clinical and MR follow-ups were conducted every 3 months in the first 2 years after treatment and every 6 to 12 months thereafter.
Imaging studies were retrospectively reviewed by two neuroradiologists (with 3 and 20 years experience), and final decisions on diagnosis, tumor location, volume, component, signal change, enhancement characteristics, and the location and stage of the hematoma were reached by consensus. The tumor was defined as “solid” when <25% of its volume was cystic; “cystic” when >75% of its volume was cystic; and “mixed” when between 25% and 75% of its volume was cystic. Intratumoral hemorrhaging was diagnosed from clinical and MR findings and/or pathological examination. On the basis of MR gradient-echo imaging or susceptibility-weighted imaging, intra-THs were classified as either intracystic or solid. Intracystic hemorrhage was indicated by blood inside cystic structures, including fluid-fluid levels within smooth-in-contour enhanced cyst walls (Figure 1). Solid hemorrhage was indicated by hematoma within the solid portion of the tumor (Figure 2). Hemorrhage was classified according to temporal change on T1-/T2-weighted images into five stages: hyperacute (T1 iso, T2 iso to high), acute (T1 iso, T2 low), early subacute (T1 high, T2 low), late subacute (T1 high, T2 high), and chronic (T1 and T2 low).7 If the MR signals of hemorrhage stage were mixed, the correlation of the most recent stage with clinical symptoms would be analyzed. The volume of hematomas and tumors was manually traced in each brain MRI slice and was calculated by multiplying the thickness of the slice by the areas of the hematoma and tumor, respectively. If there were multiple sites of hemorrhage, the total volume of the hematomas would be calculated by summing the volumes of hemorrhage at these sites. The total hematoma volume was also expressed relative to total tumor volume for better comparison between different tumor sizes and different ages of the children.8 We used the hematoma volume to tumor volume ratio (HTVR) for evaluating the functional and survival outcomes.
2.3. Statistical analysis
The statistical analysis was performed using SPSS for Windows (version 18). For univariate analysis, the nonparametric methods and χ2 test were used to assess differences in categorical variables. Receiver operating characteristic (ROC) curves were used to explore the characteristics of diagnostic tests by graphing the false positive rate (1-specificity) on the horizontal axis and the true positive rate (sensitivity) on the vertical axis with various cutoff values. A p value of 0.05 was considered to indicate a statistically significant difference among the test populations.
3.1. Patients and clinical findings
The demographic features of the 17 PIGCT patients with TH are shown in Table 1. Their mean age at diagnosis was 11.1 years (11.1 ± 4.3 [4-19]). Their diagnoses included germinoma (n = 7 [41.2%]), mixed GCT (n = 8 [47%]), yolk sac tumor (n = 1 [5.9%]), and choriocarcinoma (n = 1 [5.9%]).
Acute symptoms, such as signs of acutely increased intracranial pressure, were observed in 10 (58.8%) patients, and nonemergent symptoms, such as diabetes insipidus or lethargy, were observed in 7 patients.
The mean follow-up period was 5.6 ± 4.2 (0.8-15.2) years. The overall survival rate and 5-year survival rate for the 17 patients were 11/17 (64.7%) to date and 12/17 (70.6%), respectively. Within a mean period of 3.3 ± 2.9 (0.8-8.8) years, four patients died from tumor recurrence, one from tumor progression, and one from a radiation-related secondary tumor. The functional outcomes of all 17 patients were characterized as either poor in 9 (52.9%) or favorable in 8 (47.1%).
3.2. MR findings
The MR findings of the 17 PIGCT patients with TH are presented in Table 1. The average interval between symptom onset and MR examination was 5.9 days (5.9 ± 4.0 [2-14]). Subarachnoid seeding or metastasis was noted initially in six cases (35.3%). The mean volumes of the tumors and hematomas were 23 cm3 (23.0 ± 15.7 [1.8-53.7]) and 6.5 cm3 (6.5 ± 7.7 [0.1-25.5]), respectively. The mean HTVR was 26.4% (26.4 ± 26.4 [4.8-88.6]).
Hematomas were intracystic for nine patients (52.9%) and within solid tumor for eight (47.1%). In one case, the TH extended to the solid tumor surface and entered the ventricle. Two of the 17 (11.8%) patients had multiple stages of hematoma, which indicated repeated bleeding. The mean tumor volume, mean hematoma volume, and HTVR were 18.7 cm3, 4.5 cm3, and 22.6% for the germinoma group and 26.0 cm3, 7.9 cm3, and 29.1% for the NGGCT group and were slightly but not statistically significantly larger in the NGGCT group (p = 0.380, 0.360, and 0.631, respectively).
3.3. Statistical analysis
3.3.1. Functional and survival outcomes
The statistical analysis of factors potentially affecting functional and survival outcomes is shown in Table 2.
Regarding clinical functional outcome, the poor outcome group had significantly larger mean hematoma volume (10.4 ± 8.9 versus 2.2±1.8 cm3 in the favorable group; p = 0.021) and mean HTVR (42.3 ± 27.8% versus 8.5 ± 3.9%; p = 0.001).
Regarding the survival outcome, the deceased group had significantly larger mean HTVR (46.0 ± 31.5% versus 15.7 ± 16.1% in the living group; p = 0.016).
3.3.2. ROC curve analysis
The results of ROC curve analysis for predicting poor outcome are shown in Table 3. For hematoma volume, the estimated area under the curve was 0.83 (p = 0.021) for predicting the poor functional outcome (mRS ≧ 3), and the cutoff value was estimated to be 5.6 cm3, with 67% sensitivity and 100% specificity. The areas under the curve for HTVR as an indicator of poor functional and survival outcome were 0.97 (p = 0.001) and 0.86 (p = 0.012); the cutoff value, sensitivity, and specificity were 16.8%, 89%, and 100% for the patients with mRS≧3 and 19.3%, 83%, and 82% for the deceased group, respectively.
The incidence of spontaneous bleeding of PIGCT has been rarely reported, and it is in the range of 9.4% to 14.3%.9,10 In the study by Liang et al,10 hemorrhage was identified in the tumors of 13 NGGCT patients, whereas none was observed in the tumors of 19 germinoma patients. We found 17 cases of PIGCT with TH (7 with germinomas and 10 with NGGCTs) in a total of 116 cases of PIGCT (14.7%). Although NGGCTs (compared with germinomas) had larger tumor and hematoma volumes, and a larger percentage of them were hemorrhagic, the difference in size was not statistically significant. This result suggested that spontaneous bleeding can occur in both germinomas and NGGCTs.
Patients with germinomas have better prognosis than those with NGGCTs.4,11 The 5-year survival rate of patients with pure germinomas ranges from 89.2% to 93.3%.11,12 NGGCTs are less radiosensitive than germinomas, and the 5-year survival rates are lower, that is, 50% to 70.6%.11,12 In our studies, the 5-year survival rate was 85.7% for germinoma patients with TH and 60% for NGGCT patients with TH. The survival outcomes of both groups were similar to that reported in the literature. These findings suggested that TH by itself does not significantly influence survival. Consequently, we looked at other potential outcome predictors, such as hematoma volume.
It was unclear in the past whether TH influences the functional and survival outcomes of patients with PIGCT. In this study, functional and survival outcomes were significantly worse for patients with larger HTVR. Patients with larger hematoma volume also had worse functional outcomes. Poor functional outcome was predicted when hematoma size was >5.6 cm3, or mean HTVR was >16.8% (Table 3). These findings suggested that the hematoma volume had a significant influence on the outcomes of PIGCT patients with TH. Therefore, we recommend using hematoma volume and HTVR for predicting functional and survival outcomes in these patients. These results might be explained by acute hematoma expansion causing irreversible direct damage to the adjacent brain tissue, or tumor-induced compression of the brain causing acute increased intracranial pressure (IICP) and obstructive hydrocephalus.13 The larger the hematoma volume, the greater its effect on brain parenchyma and acute IICP. Repeated tumor bleeding might also enlarge hematoma volume and aggravate brain damage. Another possible mechanism affecting the outcome of patients with TH is injury secondary to hematoma-related inflammatory changes. Free radicals and cytokines involved in this inflammatory process can injure the brain tissue adjacent to hematomas.14 We suggested that these patients should receive aggressive treatments, such as surgical intervention or irradiation as soon as possible.
The peak incidence of PIGCT is around 10 to 12 years of age.15 In our studies, the mean age was 11.1 years. The previously reported male predominance of GCTs16 was also observed here. In our studies, TH was found in tumors located in pineal gland, suprasellar region, and basal ganglia. Hence, there is no age, sex, or tumor location predilection for spontaneous TH. That said, some clinical features of PIGCT with HT were clearly different from those of without TH (Table 1). For example, diabetes insipidus was only noted in two patients (11.8%) of this study.
Although there was no statistical significance (p = 0.102), we found that intracystic hematomas were smaller than solid hematomas. We hypothesize that intracystic hematomas are structurally confined, limiting their adjacent extension and size expansion. These confined intracystic hematomas might resolve with tumor control by irradiation. In contrast, vascular hematomas within solid tumors have no defined cystic wall barrier. This intratumoral hematoma may expand even beyond the tumor surface and have a significant mass effect possibly justifying early surgical intervention.
The limitations of this retrospective study included small number of patients, diverse MR patterns of TH, and lack of a single therapeutic strategy. These limitations made statistical analysis difficult. We suggest a prospective, multicenter study to evaluate more cases. A study using comprehensive imaging techniques, such as MR perfusion imaging, may be beneficial in evaluating the disease process before and after therapy.
In conclusion, our study demonstrated that (1) functional and survival outcomes were significantly worse for patients with larger HTVRs than those with smaller ones; (2) tumors with hematomas >5.6 cm3 or mean HTVR >16.8% increased risk of poor functional outcome (mRS ≧ 3); and (3) mean HTVR >19.3% predicted poor survival outcome after treatment. In PIGCT patients with TH, early and aggressive treatment (such as radiotherapy) is recommended to improve the clinical outcomes of those with large hematoma volume and large HTVR.
This study was supported by Taipei Veterans General Hospital (V106C-197) and Ministry of Science and Technology (105-2314-B-075-027-MY2).
We also thank Dr. Z. Sean Juo for English editing of this manuscript.
1. Makino K, Nakamura H, Yano S, Kuratsu J. Population-based epidemiological study of primary intracranial tumors in childhood.Childs Nerv Syst2010261029–34
2. Jennings MT, Gelman R, Hochberg F. Intracranial germ-cell tumors: natural history and pathogenesis.J Neurosurg198563155–67
3. Sawamura Y, Ikeda J, Shirato H, Tada M, Abe H. Germ cell tumours of the central nervous system: treatment consideration based on 111 cases and their long-term clinical outcomes.Eur J Cancer199834104–10
4. Matsutani M, Sano K, Takakura K, Fujimaki T, Nakamura O, Funata N, et al. Primary intracranial germ cell tumors: a clinical analysis of 153 histologically verified cases.J Neurosurg199786446–55
5. Lv XF, Qiu YW, Zhang XL, Han LJ, Qiu SJ, Xiong W, et al. Primary intracranial choriocarcinoma: MR imaging findings.Am J Neuroradiol2010311994–8
6. Aarsen FK, Paquier PF, Reddingius RE, Streng IC, Arts WF, Evera-Preesman M, et al. Functional outcome after low-grade astrocytoma treatment in childhood.Cancer2006106396–402
7. Bradley WG Jr.. MR appearance of hemorrhage in the brain.Radiology199318915–26
8. Jordan LC, Kleinman JT, Hillis AE. Intracerebral hemorrhage volume predicts poor neurologic outcome in children.Stroke2009401666–71
9. Wang Y, Zou L, Gao B. Intracranial germinoma: clinical and MRI findings in 56 patients.Childs Nerv Syst2010261773–7
10. Liang L, Korogi Y, Sugahara T, Ikushima I, Shigematsu Y, Okuda T, et al. MRI of intracranial germ-cell tumours.Neuroradiology200244382–8
11. Carlos Chung KH, Owler BK, Dexter M, Chaseling R. Paediatric germ cell tumours of the central nervous system: results and experience from a tertiary-referral paediatric institution in Australia.J Clin Neurosci201320514–9
12. Foo AS, Lim C, Chong DQ, Tan DY, Tham CK. Primary intracranial germ cell tumours: experience of a single South-East Asian institution.J Clin Neurosci2014211761–6
13. Vinchon M, Rekate H, Kulkarni AV. Pediatric hydrocephalus outcomes: a review.Fluids Barriers CNS20129
14. Hanley DF, Thompson RE, Muschelli J, Rosenblum M, McBee N, Lane K, et al. Safety and efficacy of minimally invasive surgery plus alteplase in intracerebral haemorrhage evacuation (MISTIE): a randomised, controlled, open-label, phase 2 trial.Lancet Neurol2016151228–37
15. Echevarria ME, Fangusaro J, Goldman S. Pediatric central nervous system germ cell tumors: a review.Oncologist200813690–9
16. Smith AB, Rushing EJ, Smirniotopoulos JG. From the archives of the AFIP: lesions of the pineal region: radiologic-pathologic correlation.Radiographics2010302001–20